using the asphalt pavement analyzer to assess rutting ...docs.trb.org/prp/12-3689.pdf · the...
TRANSCRIPT
Using the Asphalt Pavement Analyzer to Assess Rutting Susceptibility of 1
HMA Designed for High Tire Pressure Aircraft 2
3
Submission date: November 14, 2011 4
5
Authors: 6
John F. Rushing (corresponding author) 7 U.S. Army Engineer Research and Development Center 8
CEERD-GM-A 9 3909 Halls Ferry Road 10
Vicksburg, MS 39180-6199 11 Phone: (601) 634-3577 12
Fax: (601) 634-3020 13 E-mail: [email protected] 14
15 16
Dallas N. Little, Ph.D. 17 Texas A&M University 18
CE/TTI 603 19 3136 TAMU 20
College Station, TX 77843-3136 21 Phone: (979) 845-9847 22
Fax: (979) 845-6156 23 E-mail: [email protected] 24
25 Navneet Garg, Ph.D. 26
Federal Aviation Administration 27 Airport Technology R&D Branch 28
William J. Hughes Technical Center 29 Atlantic City Intl. Airport, NJ 08405 30
Phone: (609) 485-4483 31 Fax: (609) 485-4845 32
E-mail: [email protected] 33 34 35
36
37
Word Count 38 Total: 7,496 39 Text: 5,746 40
Tables and Figures (7*250): 1,750 41 42
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 2
ABSTRACT 1
Hot Mix Asphalt (HMA) laboratory mix design is intended to determine the proportion of aggregate and 2
binder that, when mixed and compacted under a specified effort, will withstand anticipated loading 3
conditions. Current mix design procedures using the Superpave Gyratory Compactor (SGC) rely on 4
engineering properties and volumetrics of the compacted mixture to assure reliable performance; 5
however, a definitive performance test does not exist. The Asphalt Pavement Analyzer (APA) is 6
evaluated as a tool to assess HMA mixtures designed to perform under high tire pressure aircraft 7
following Federal Aviation Administration (FAA) specifications. The APA used in this study was 8
specially designed to test simulated high tire pressures of 250 psi, common for aircraft. Thirty-three 9
HMA mixtures are included in the study. Each was designed using the SGC according to preliminary 10
criteria being developed by the FAA. The study included some mixtures that contained excessive 11
percentages of natural sand and that do not meet FAA criteria. These mixtures are included to provide 12
relative performance for mixtures expected to exhibit premature rutting. Asphalt Pavement Analyzer 13
testing using the high tire pressure APA resulted in rapid failure of HMA specimens compared to 14
traditional APA testing at lower pressures. Data were analyzed with the focus of providing acceptance 15
recommendations for mixtures to support high tire pressures. A preliminary 10 mm rut depth criterion 16
after 4,000 load cycles is recommended. 17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 3
INTRODUCTION 1
Rutting is a primary load-related distress in airport pavements subjected to high tire pressure aircraft 2
loads. Hot Mix Asphalt (HMA) for airport pavements has historically been designed using materials and 3
compaction requirements that can withstand these loading conditions. The Marshall design procedure has 4
been used by the Federal Aviation Administration (FAA) for airport pavement. The design procedure 5
includes meeting aggregate and binder requirements along with the laboratory mix design volumetric 6
requirements to produce acceptable mixtures. Marshall stability and flow test values provide an 7
empirical, threshold metric to help insure a stable mixture under traffic, but this empirical test cannot 8
assess performance. Performance can be assessed by either a damage model that is based on sophisticated 9
testing of engineering and material properties or tests that mimic the traffic loads encountered in the field. 10
A new mix design procedure being implemented by the FAA allows the use of the Superpave 11
Gyratory Compactor (SGC) in lieu of Marshall impact compaction to compact specimens during the mix 12
design process (1). This new procedure includes the same aggregate and binder requirements and the 13
same volumetric requirements as those that are currently part of the Marshall protocol, but no 14
performance test is included in the protocol. Studies are currently being conducted to identify a 15
companion performance test for the new mix design procedures, but one has not yet been adopted. 16
The Asphalt Pavement Analyzer (APA) has been successfully used by several agencies as a test 17
to determine HMA mixture suitability and is one of the most widely accepted laboratory accelerated 18
wheel trafficking devices available and as one that attempts to mimic the action of a moving and heavily 19
loaded, high tire pressure wheel of the type that simulates airfield traffic. Tests that simulate or mimic 20
traffic action are often used in lieu of traditional laboratory testing because the action of a moving wheel 21
load in which there is a rotation of principal stresses and a transition from compression to extension is 22
difficult to recreate with even sophisticated triaxial laboratory tests (2). For this reason, empirical tests 23
such as the APA offer an alternative approach, which is favored by many. 24
The APA was first manufactured in 1996 (3). It places a loaded wheel on a pressured linear tube, 25
centered on top of the specimen. The wheel is tracked in a forward and backward linear motion across 26
the sample(s), resulting in plastic deformation, or rutting. The APA can be used to test either cylindrical 27
or beam specimens. 28
Several studies have attempted to correlate APA laboratory rutting with field rutting. Williams 29
and Prowell (4) found that the APA test results correlated well with field results, as did a number of state 30
departments of transportation (5-8). However, as the WesTrack Forensic Team Study demonstrated, a 31
direct relationship between lab rutting and field rutting does not exist. For example the WesTrack study 32
showed that a laboratory rut depth of 6 mm after 8,000 cycles represented a field rut depth of 12.5 mm 33
(9). 34
Choubane, Page, and Musselman (7) reported appropriate ranking of mixes according to field 35
performance. However, their study indicated APA test variability was statistically significant from test to 36
test and among locations during a test. Recommendations from the study cautioned on using the APA as 37
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 4
a pass or fail criterion based on the three mixes tested. AASHTO TP 63-07 indicated precision and bias 1
statements have yet to be determined for APA testing. 2
OBJECTIVE 3
The objective of this study was to evaluate the ability of the APA to serve as a performance test to assess 4
the potential for rutting under high tire pressure aircraft during the design of hot asphalt mixtures. HMA 5
mixtures carefully designed to exhibit a wide range of rutting potential were included in the study. The 6
ability of the APA to differentiate among these mixtures and its applicability as a mix design and/or 7
quality assurance test were investigated. The actual rutting susceptibility of the mixtures tested under 8
field conditions was not evaluated; however, mix variables known to contribute to rutting were controlled 9
in order to provide relative performance correlations. 10
TEST PROCEDURE 11
The APA used in this study was specifically designed to simulate high tire pressures associated with 12
aircraft. Two such high pressure APAs are known to exist in July 2011: one at the FAA William J. 13
Hughes Technical Center in Atlantic City, New Jersey, and one at the Air Force Research Laboratory at 14
Tyndall Air Force Base, Florida. Testing in this study was conducted at the FAA test facility. Mix 15
design and specimen preparation were accomplished at the U.S. Army Engineer Research and 16
Development Center (ERDC), Waterways Experiment Station in Vicksburg, Mississippi. 17
An APA tube or hose pressure of 250 psi under a wheel load of 250 lb was used for testing. The 18
test temperature was 64oC as was the high temperature PG grade for the binder used in all mixtures. 19
Cylindrical asphalt concrete specimens with a target air void content of 3.5% were prepared and tested. 20
The air void content was selected as the midpoint of the allowable range in the FAA mix design 21
procedure. Two replicate specimens were tested for each mix. The APA reports the average rut depth of 22
the two specimens. Thirty-three mixtures were included in the study. The mixtures used one neat binder 23
and three aggregate types. Some mixtures included natural sand. 24
Cyclic loads were applied by the APA at a rate of one cycle per second. The terminal rut depth of 25
the specimens was set at 12 mm after 8,000 cycles; however, the test was terminated when the 12 mm rut 26
depth was achieved if this occurred before 8,000 cycles. Once one of the two specimens reached terminal 27
rut depth, the test was stopped. However, since the APA reports the average rut depth for the two 28
specimens, some average rut depths were less than 12 mm. 29
MATERIALS 30
The asphalt binder used in this study was obtained from Ergon Asphalt and Emulsions, Inc. Tests by the 31
distributor indicated the binder graded as a PG 64-22 and had a specific gravity of 1.038. Supplier-32
recommended mixing and compaction temperatures for the binder were 310oF (154
oC) and 290
oF 33
(145oC), respectively, based on binder viscosity. 34
Aggregates used in this study consisted of material stockpiles at the ERDC collected for a 35
previous study which evaluated the use of the SGC for airport mix design (9). These aggregates included 36
limestone, granite, and chert gravel. The limestone aggregate was from a Vulcan Materials quarry in 37
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 5
Calera, Alabama. The granite aggregate was from Granite Mountain Quarries in Little Rock, Arkansas. 1
The chert gravel aggregate was from Green Brothers Gravel Company in Copiah County, Mississippi. 2
Additionally, some mixtures were blended with selected percentages of natural sand, which was locally 3
purchased from Mississippi Materials Corporation in Vicksburg, Mississippi. 4
Each aggregate type was comprised of multiple stockpiles that were blended to meet the target 5
gradations. Selected gradations were within the allowable range of size fractions incorporated in FAA 6
specifications (11). The gradations referred to in this paper are designated as fine and coarse. Fine 7
gradations are those near the upper limits of the gradation band. Coarse gradations are those near the 8
lower limits of the gradation band. Some aggregate blends included 10 or 30 percent natural sand. These 9
gradations are characterized by a hump in the grain size distribution near the 30 to 50 sieve sizes. 10
The percentages of aggregate with at least two fractured faces were 100, 100, and 97 percent for 11
the limestone, granite, and chert gravel, respectively. The maximum percentages of flat and elongated 12
aggregates were 1.6, 1.0, and 0.3 percent for the limestone, granite, and chert gravel, respectively. 13
The fine aggregate angularity for the limestone, granite, chert gravel, and mortar sand aggregates 14
was determined by Method A of ASTM C 1252. The limestone, granite, and chert gravel aggregates had 15
a fine aggregate angularity of 47, 47, and 46 percent, respectively. The fine aggregate angularity of the 16
mortar sand was 40 percent. This value is characteristic of rounded aggregate particles and is typical for 17
natural sands (12). 18
Additional testing of the aggregates was performed using the Aggregate Imaging System (AIMS). 19
AIMS determines shape characteristics of aggregates through image processing and analysis techniques 20
(13). AIMS is a computer automated system that includes a lighting table where aggregates are placed in 21
order to measure their physical characteristics (shape, angularity and texture). It is equipped with an 22
autofocus microscope and a digital camera, and is capable of analyzing the characteristics of aggregates 23
sizes retained on the #100 sieve (0.15 mm sieve) up to aggregates retained on the 1 inch sieve (25.4 mm). 24
Texture is measured by analyzing gray scale images captured at the aggregate surface using the wavelet 25
analysis method. The surface irregularities manifest themselves as variations in gray-level intensities that 26
range from 0 to 255. Large variations in gray-level intensity mean a rough surface texture, whereas a 27
smaller variation in gray-level intensity means a smooth particle. The wavelet transform analyzes the 28
image as a two dimensional signal of gray scale intensities, and it gives a higher texture index for 29
particles with rougher surfaces. Angularity is measured using the gradient analysis method which 30
basically quantifies the change in angles along the circumference of a particle. A higher change in angle 31
means a more angular particle. Masad et al. (13) gives detailed background information about AIMS 32
operations and analysis methods. Six size fractions for each of the three aggregate types were tested 33
using AIMS. These fractions included aggregates retained on 1/2 in., 3/8 in., #4, #8, #16, and #30 U.S. 34
standard sieves during a washed sieve analysis. Table 1 shows the average results for angularity and 35
texture indices measured by AIMS. 36
37
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 6
TABLE 1 Aggregate data from AIMS analysis 1
Sieve Size
Angularity Texture
Limestone Granite Chert Gravel Limestone Granite Chert Gravel
1/2 in. 2607 3200 2721 359 535 153
3/8 in. 2668 3167 2912 345 493 164
#4 2841 3461 2960 274 362 130
#8 3162 3709 3212
#16 3164 3907 3348
#30 3176 3876 3282
Texture is only measured on coarse aggregate
2
MIX DESIGN 3
A Pine Instruments Company model AFGC125X gyratory compactor was used in the mix designs during 4
this study to produce cylindrical asphalt concrete specimens with a diameter of 150 mm at a target height 5
of 115 mm. Compaction was performed using a ram pressure of 87 psi (600 kPa) and an internal angle of 6
gyration of 1.16o ± 0.02
o. Asphalt mixtures were compacted to 70 gyrations at a rate of 30 revolutions per 7
minute. Seventy gyrations is recommended for Ndesign for HMA mixtures designed for high tire pressure 8
aircraft (9), but this is currently being evaluated by the FAA. Three different binder contents were used 9
for the mix designs, in increments of 0.5 % binder. Three replicate specimens were compacted at each 10
binder content. The mix designs were able to bracket the design binder content using only three trial 11
percentages because of previous experience with these materials (9). The air void content and voids in 12
mineral aggregate (VMA) were determined according to Asphalt Institute MS-02 procedures (14). The 13
design or optimal binder content was selected as the binder content that resulted in a compacted specimen 14
having 3.5% air voids. 15
Specimens produced for APA testing were compacted using the design binder content at a target 16
height of 75 mm. The mass of binder and aggregate was proportioned so that the compacted specimen 17
height would be near the target value. Reducing the mass of the material in the mold was expected to 18
have some influence on the volumetric properties. The smaller sample size, along with inherent specimen 19
variability, resulted in some specimens having higher or lower air voids than the target value of 3.5%. 20
Table 2 shows the aggregate types, maximum size, and relative gradation as well as the 21
percentage of natural sand, the binder content, and the air void content for the APA test specimens. The 22
mix designation is also listed and is used to identify these mixtures in this paper. 23
24
25
26
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 7
1
TABLE 2 APA test specimen characteristics 2
Results from Superpave Mix Design at 70 Gyrations (3.5% Air Voids)
Aggregate
Type
Maximum
Aggregate
Size Gradation
Percentage
of Mortar
Sand
Optimum
Binder
Content
Air
Voids 1
Air
Voids 2
Mix
Designation
Granite
1/2"
Fine
0 6.7 3.73 3.68 1/2 FGN
10 6.8 3.07 3.13 1/2 FGN10
30 7.2 5.49 5.95 1/2 FGN30
Coarse
0 6.3 3.76 3.95 1/2 CGN
10 5.9 3.85 3.62 1/2 CGN 10
30 6.8 4.89 3.52 1/2 CGN 30
3/4"
Fine
0 6.2 2.37 3.02 3/4 FGN
10 6.1 2.57 0.83 3/4 FGN 10
30 7.0 2.54 3.17 3/4 FGN 30
Coarse
0 5.9 4.70 6.30 3/4 CGN
10 4.9 4.27 4.94 3/4 CGN 10
30 7.1 5.03 4.56 3/4 CGN 30
Limestone
1/2"
Fine
0 6.1 4.26 4.11 1/2 FLS
10 5.2 3.69 3.50 1/2 FLS 10
30 6.9 2.90 2.91 1/2 FLS 30
Coarse
0 5.5 4.65 5.13 1/2 CLS
10 5.0 4.05 3.58 1/2 CLS 10
30 6.1 4.00 4.15 1/2 CLS 30
3/4"
Fine
0 5.7 4.53 4.58 3/4 FLS
10 4.8 3.90 4.40 3/4 FLS 10
30 5.9 4.06 3.63 3/4 FLS 30
Coarse
0 5.4 2.80 3.17 3/4 CLS
10 5.4 2.04 2.50 3/4 CLS 10
30 5.7 3.79 3.82 3/4 CLS 30
Chert
Gravel
1/2" Center
0 6.8 3.45 3.34 1/2 FGV
10 6.2 2.50 2.28 1/2 FGV 10
30 6.8 2.79 2.83 1/2 FGV 30
3/4"
Fine
0 6.8 3.74 3.73 3/4 FGV
10 5.9 2.90 2.89 3/4 FGV 10
30 7.1 2.43 2.45 3/4 FGV 30
Coarse
0 6.4 3.70 3.90 3/4 CGV
10 5.3 3.34 3.16 3/4 CGV 10
30 6.6 2.76 2.75 3/4 CGV 30
3
4
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 8
1
APA RESULTS 2
General 3
The APA records the average rut depth of the two specimens with each load cycle into a Microsoft Excel 4
spreadsheet. The average rut depth was plotted versus the number of load cycles to produce a curve of 5
accumulated rutting. Figure 1 shows the number of load cycles versus APA rut depth for all 33 mixtures. 6
As previously stated, the test was ceased after one of the two specimens reached the terminal rut depth of 7
12 mm. The values shown in Figure 1 are the average rut depth of the two specimens. The mix 8
designation is not included on this figure because of the large number of data sets. Data from Figure 1 are 9
extracted in subsequent figures with mix designations included for analysis. 10
11
FIGURE 1 APA results for all mixtures 12
During APA testing, a more rapid rutting rate took place during the initial load cycles. After 13
about 1 mm of rut depth, specimen behavior was observed to be more closely linked to mixture 14
characteristics as evidenced by variable rates of rut depth accumulation. Some mixtures had a slow rate 15
of rut depth accumulation while others failed very quickly. The rate of rut depth accumulation appeared 16
to become somewhat linear after approximately 2 mm of rutting. The rutting behavior in the APA 17
0
2
4
6
8
10
12
14
0 1000 2000 3000 4000 5000 6000 7000 8000
Ru
t D
epth
(mm
)
Load Cycles
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 9
followed the same general pattern as commonly observed in creep and repeated loading experiments with 1
a primary and secondary flow. Tertiary flow was not observed during the experiments. 2
Although mix designations are not shown in Figure 1, analysis of the data indicates general 3
trends in mix variables that impact the rate of rut depth accumulation. These mix variables are 4
investigated in the following paragraphs. 5
An analysis was conducted to determine the effect of natural sand on the APA results. Figure 2 6
shows the APA results for mixtures containing no natural sand. These mixtures, as expected, were among 7
the best performers in the APA test. Incorporating natural sand promotes rutting in HMA (15, 16). For 8
mixtures containing no natural sand, the accumulation of rut depth is related to the aggregate type. 9
Mixtures containing crushed chert gravel aggregate rutted much more quickly than the other mixtures. 10
The crushed gravel meets the FAA requirements for the mass percentage of aggregate particles having at 11
least two fractured faces (70% for coarse aggregate). However, these aggregates also have low levels of 12
angularity and relatively smooth texture. Interparticle friction, although not directly measured, is 13
accepted to be lower for chert gravel than for quarried aggregate. In addition, chert gravel mixtures 14
commonly have a higher VMA than quarried aggregate mixtures, resulting in more binder required to 15
compact mixtures to equivalent air void content. 16
Mixtures containing crushed limestone aggregate performed best in the APA test. In general, the 17
crushed granite mixtures rutted more quickly than the crushed limestone mixtures. The crushed granite 18
mixtures, on average, had a higher design binder content than the crushed limestone mixtures. The 19
differences in design binder content or binder demand are likely to be influenced by factors such as 20
aggregate shape, texture, and breakdown during compaction or mixture VMA. 21
22
23
24
25
26
27
28
29
30
31
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 10
1
2
3
FIGURE 2 APA results for mixtures containing no natural sand 4
Figure 3 shows the APA results for mixtures containing 10% natural sand. These mixtures rutted 5
more quickly than those containing no natural sand. Once again, mixtures produced with crushed gravel 6
rutted more quickly than other mixtures. Similarly, mixtures produced with crushed limestone performed 7
best by demonstrating the greatest resistance to rutting. 8
0
2
4
6
8
10
12
14
0 1000 2000 3000 4000 5000 6000 7000 8000
Ru
t D
epth
(m
m)
Load Cycles
1/2 GV
3/4 FGV
3/4 CGV
1/2 FLS
1/2 CLS
3/4 FLS
3/4 CLS
1/2 FGN
1/2 CGN
3/4 FGN
3/4 CGN
1/2 GV 1/2 FGN3/4 FGV
1/2 CGN
3/4 CGV
1/2 FLS
1/2 CLS
3/4 CGN
3/4 FGN
3/4 FLS
3/4 CLS
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 11
1
FIGURE 3 APA results for mixtures containing 10% natural sand 2
Some mixtures contained 30% natural sand, a higher percentage of natural sand than is allowed 3
by FAA specifications (maximum of 15%). The mixtures were included in the analysis because they 4
were expected to rut quickly and could provide guidance for performance threshold levels within the 5
specifications. All mixtures containing 30% natural sand failed quickly (less than 1,500 cycles) in the 6
APA. 7
In general, the mixture variable with greatest influence on APA test results was the percentage of 8
natural sand. High percentages of natural sand caused premature failure in the APA. Additionally, 9
aggregate type influenced the APA test results. Mixtures of chert gravel rutted more quickly than 10
mixtures with granite or limestone aggregate. 11
Statistical Considerations 12
Analyses were performed to evaluate the impact of mixture variables on rut depth accumulation 13
considering statistical significance. All analyses were performed using SigmaStat® software at a 95% 14
confidence level. The analysis of variance (ANOVA) procedure, including the Tukey test for all pairwise 15
comparison, was used to evaluate data sets. The number of load cycles to reach an average rut depth of 16
0
2
4
6
8
10
12
14
0 1000 2000 3000 4000 5000 6000 7000 8000
Ru
t D
epth
(m
m)
Load Cycles
1/2 GV10
3/4 FGV10
3/4 CGV10
1/2 FLS10
1/2 CLS10
3/4 FLS10
3/4 CLS10
1/2 FGN10
1/2 CGN10
3/4 FGN10
3/4 CGN10
3/4 CLS 10
1/2 FLS 10
3/4 FLS 10
3/4 FGN 10
1/2 CLS 10
3/4 CGN 10
3/4 CGV 10
3/4 FGV 10
1/2 CGN 101/2 GV 10
1/2 FGN 10
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 12
10 mm was used as a metric to quantify mixture behavior in these analyses. A rut depth of 10 mm was 1
selected because it is near the maximum average rut depth reported in the APA samples. 2
The statistical analyses considered the following factors: (1) the influence of aggregate type, i.e., 3
limestone, chert gravel or granite; (2) the influence of shape and textural factors (angularity, surface 4
texture and flat and elongated shape factors); and (3) the impact of the presence and amount of field or 5
natural sand (uncrushed). First, the influence of aggregate type on APA results was investigated. Granite, 6
limestone, and chert gravel mixtures containing no natural sand require statistically different numbers of 7
load cycles to reach 10 mm of rutting in the APA. The average number of load cycles required to reach 8
10 mm rutting for these respective mixtures was 6,530; 8000; and 1,740. It is important to note that the 9
value of 8,000 cycles is the terminal test value and no data were extrapolated. 10
Further analyses on the influence of aggregate type on APA rutting were performed on the AIMS 11
data. The indices of shape and angularity were investigated and considered to be the aggregate 12
characteristics most closely related to rutting in HMA. Aggregate angularity was measured for three sizes 13
of coarse aggregate and three sizes of fine aggregate for each aggregate type using AIMS. Coarse and 14
fine aggregate were compared independently using ANOVA. Angularity was not found to be statistically 15
different among the three size fractions within a specific aggregate type. The difference between the 16
coarse aggregate fraction of limestone and chert gravel aggregates was not statistically significant. For 17
each coarse size fraction, the granite aggregate was more angular than limestone. Granite was statistically 18
more angular than chert gravel aggregate on ½ in. and #4 sieve sizes, but not the 3/8 in. sieve. Similarly, 19
no statistically significant difference in angularity was detected between limestone and chert gravel fine 20
aggregate. All size fractions of fine granite aggregate were statistically more angular than all size 21
fractions of limestone or chert gravel aggregate. 22
Aggregate texture is only measured on coarse aggregate by AIMS. Statistical analyses of AIMS 23
texture data indicate all sizes of chert gravel have a statistically lower texture index than any size fraction 24
of limestone or granite aggregate. For all equal size fractions, granite aggregate has a higher texture than 25
limestone aggregate. 26
In summary, the AIMS data rank the aggregates the same by both angularity and texture, with 27
granite having the highest indices and chert gravel having the lowest indices, although the angularity of 28
limestone and chert gravel are very similar. Higher angularity and texture indices are expected to result in 29
greater rutting resistance. On this basis, the degree of angularity and texture are consistent in terms of 30
identifying chert gravel mixtures as the most rut susceptible, but inconsistent in predicting granite as the 31
most rut resistant mixtures. APA results indicate aggregate texture may be a better indicator of rutting 32
resistance than angularity. The lower resistance of the granite mixtures to APA rutting compared to 33
limestone mixtures likely results from higher design binder contents. In fact, a Pearson’s product moment 34
correlation analysis considering the influence of aggregate angularity and texture, mixture binder content, 35
and specimen air void content on APA rutting indicated aggregate texture and binder content were the 36
only statistically influential variables. Higher rut resistance resulted from increased aggregate texture and 37
lower binder content. 38
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 13
The addition of neither 10% nor 30% natural sand to the mixtures impacted the rank order of 1
rutting sensitivity among the three aggregate types. For mixtures containing 10% natural sand, the mean 2
number of load cycles resulting in 10 mm rut depth was 5,518; 7,033; and 2,850 for granite, limestone, 3
and chert gravel mixtures, respectively. The trend of performance, according to the number of load 4
cycles resulting in 10 mm rutting, was the same for mixtures containing no natural sand as for mixtures 5
containing 10% natural sand. Limestone mixtures performed the best while chert gravel mixtures 6
performed the poorest. However, only limestone and chert gravel were statistically different from each 7
other. For mixtures containing 30% natural sand, the numbers of load cycles to 10 mm rut depth were 8
490, 790, and 784, respectively. Each of these mixtures failed rapidly compared to mixtures with a higher 9
percentage of crushed aggregate. 10
The effect of the percentage of natural sand contained in the mixtures on the number of load 11
cycles to develop 10 mm of rutting was analyzed independently. The addition of 10% natural sand had 12
no statistically significant impact on the number of load cycles to reach 10 mm rut depth for the granite 13
aggregate. However, granite mixtures containing 30% natural sand were statistically different from 14
mixtures containing either no natural sand or 10% natural sand. Similarly, statistical analyses of 15
limestone mixtures revealed that mixtures containing no natural sand and 10% natural sand were not 16
statistically different from each other, but both were different from mixtures containing 30% natural sand 17
in terms of number of cycles to 10 mm rut depth. The addition of field sand had no statistically verifiable 18
impact on load cycles to 10 mm rutting for chert gravel mixtures, regardless of the percentage of natural 19
sand in the mix. The effect of neither maximum aggregate size nor relative gradation was statistically 20
significant. 21
Selection of Interim Mixture Evaluation Criteria (Threshold Values) 22
In order to establish interim mixture evaluation criteria for the APA, the number of load cycles to reach 23
different rut depths was considered. Rut depths of 4 mm, 6 mm, 8 mm, and 10 mm were selected as 24
possible failure thresholds. The number of load cycles required to reach these levels could be easily 25
determined since the APA recorded the average rut depth after each load cycle. Figure 4 and 5 show the 26
number of load cycles to achieve these threshold average rut depths. In some cases, mixtures did not 27
reach an average rut depth of 8 or 10 mm before the test was terminated. A value of 8000 cycles was 28
denoted as the failure point for these mixtures, even though many more load cycles might have actually 29
been required to obtain these rut depths. 30
The data were separated into two figures identified by the percentage of natural sand in the 31
mixtures, to allow for easier interpretation by reducing the total number of data points in one figure. Only 32
data for mixtures containing no natural sand and 10% natural sand are presented. The mixtures 33
containing 30% natural sand all failed rapidly in the APA. All but one of these mixtures reached 10 mm 34
average rut depth before 1,000 cycles. 35
36
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 14
1
FIGURE 4 Number of load cycles versus target rut depths for mixtures with no natural sand 2
3
4
FIGURE 5 Number of load cycles versus target rut depths for mixtures with 10% natural sand 5
6
0
1000
2000
3000
4000
5000
6000
7000
8000
4 6 8 10
Nu
mb
er
of
Load
Cyc
les
Rut Depth (mm)
1/2 FGN
1/2 CGN
3/4 FGN
3/4 CGN
1/2 FLS
1/2 CLS
3/4 FLS
3/4 CLS
1/2 GV
3/4 FGV
3/4 CGV
1/2 FGN
3/4 CGN
3/4 CGV
1/2 GV
3/4 FGV
1/2 CGN
3/4 FGN1/2 CLS
3/4 CLS 1/2 FLS
3/4 FLS
0
1000
2000
3000
4000
5000
6000
7000
8000
4 6 8 10
Nu
mb
er
of
Load
Cyc
les
Rut Depth (mm)
1/2 FGN10
1/2 CGN 10
3/4 FGN 10
3/4 CGN 10
1/2 FLS 10
1/2 CLS 10
3/4 FLS 10
3/4 CLS 10
1/2 GV 10
3/4 FGV 10
3/4 CGV 10
3/4 CLS 10
1/2 FLS 10
3/4 FLS 10
1/2 CLS 10
3/4 FGN 10
1/2 FGN 10
1/2 CGN 10
3/4 CGN 10
1/2 GV 10
3/4 CGV 10
3/4 FGV 10
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 15
The data in Figures 4 and 5 indicate a relatively linear rate of rut depth accumulation. Primary 1
rutting (rapid accumulation) generally occurred until a rutting magnitude of about 4 mm. The tertiary 2
phase of rutting was not reached before testing ended. Mixtures containing 30% natural sand rutted faster 3
than mixtures containing either 10% or no field sand in all cases 4
To provide perspective on these data, researchers compared these results to criteria for agencies 5
that use the APA to accept HMA mixtures. These criteria were obtained from Williams et al. (3), and are 6
current as of 2005. Other criteria, later than 2005, may exist but were not found in the literature search. 7
Results from the APA test with a 250 psi hose pressure are expected to be significantly different from 8
results used to develop criteria for highway pavements. 9
The majority of agencies require APA test specimens to be fabricated with a target air void 10
content of 7%. Of the twenty-one states reported in Williams’ paper, only Alabama, Arkansas, and New 11
Jersey test specimens a different air voids content, 4% air voids, which was the design air void content. 12
Although higher air voids are more indicative of field-placed mixtures, much of the rutting at this air void 13
content may be related to densification and not shear (17). Additionally, using the design air void content 14
allows for performance testing of specimens produced from the mix design. 15
Several agencies vary the APA requirements based on the number of design equivalent single 16
axle loads (ESALs). Others have a singular criterion or were listed as still evaluating the APA. Only five 17
states included the APA in the specifications; others used the APA for comparative purposes. All 18
agencies require the application of 8,000 load cycles during testing. The average maximum allowable rut 19
depth specified is 5 mm. The lowest maximum rut depth is 3 mm (Arkansas and Delaware), while the 20
highest maximum rut depth is 10 mm (Mississippi). 21
Comparison of results from this study with the agency criteria indicated that the high tire pressure 22
APA is much more damaging than the traditionally used APA. After application of 8,000 load cycles 23
with 250 psi, no mixtures rutted less than 6 mm. Only four mixtures rutted less than 8 mm, while only 24
nine mixtures rutted less than 10 mm. Aside from the mixtures containing 30% natural sand, all mixtures 25
met or exceeded all FAA criteria for HMA mix design. 26
Since the APA is an empirical test that mimics loading, the most valid way to establish threshold 27
values are through correlations with field studies. For example, the criterion recommended by the 28
National Center for Asphalt Technology (NCAT) was less than 8.2 mm after 8,000 cycles at the location 29
high-temperature PG grade (18). This criterion was developed through correlation of field data at the 30
NCAT test track. The traffic used in the field test was 10,000,000 ESALs. These types of correlations 31
indicate the large magnitude of damage each APA cycle induces compared to actual traffic. Ultimately, 32
the rutting threshold established for the heavy wheel load (high tire pressure) APA must be correlated to 33
field results. However, in the interim it is important to establish a viable and realistic interim threshold 34
acceptance rutting relationship between number of passes of the APA and rut depth. In order to do this, 35
we considered the following factors: (1) the number of applications of aircraft loading relative to highway 36
truck loading, (2) the tire pressure of airfield traffic relative to highway truck traffic, and (3) established 37
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 16
correlations of the type summarized in this discussion. Based on these considerations we also considered 1
the results of the mixtures that were tested and the historical performance of these mixtures in airfield 2
situations. The following paragraphs summarize the reasons for establishing the interim criteria. 3
Either 8 or 10 mm of rutting could be used as a criterion for maximum allowable APA rut depth 4
when tested at 250 psi for airport pavements. The number of load cycles at which this level is reached 5
has to be selected to correspond with mixture performance. Using the higher value will still produce a 6
test that can be run in a reasonable timeframe (around 1 hr) and has the ability to differentiate among 7
mixtures. 8
To eliminate the mixtures containing 30% natural sand in this study, the criterion would need to 9
be a maximum of 10 mm rut depth after 1,000 load cycles. However, this value is still inclusive of some 10
mixtures. Although field performance is unknown, the mixtures containing chert gravel rutted 11
significantly more quickly than mixtures containing quarried aggregate. A criterion of less than 10 mm 12
APA rut depth after 3,000 cycles would eliminate all but one of the chert gravel mixtures. This criterion 13
would also eliminate one granite mixture containing 10% natural sand. 14
Based on the data from this study, a reasonable criterion for airport HMA designed for high tire 15
pressure aircraft is less than 10 mm APA rut depth after 4,000 cycles when tested using 250 psi hose 16
pressure. This criterion would eliminate 18 of the 33 mixtures in this study. However, 11 of the 18 17
mixtures that failed were not acceptable mixtures because they contained excessive natural sand. Of the 18
other seven failed mixtures, five were chert gravel mixtures that may not be commonly used in airport 19
HMA. Mixture improvements can be made by adjusting the gradation or binder content, and a key 20
advantage of the APA is that the test can be performed in slightly more than one hour and would be easily 21
implementable for HMA mix design. 22
CONCLUSIONS 23
There is a pressing need for a performance test to accompany HMA mix design for airport pavements. 24
Although the Marshall design method uses an empirical index test, a new design method using the SGC 25
relies only on volumetric properties of the compacted mixture for acceptance. This study investigated the 26
suitability of the APA as a test for characterizing HMA for airport pavements subjected to high tire 27
pressure aircraft. From this study we conclude the following: 28
The most significant factor influencing APA rutting is excessive natural sand (30%). Mixtures 29
containing excessive natural sand achieved 10 mm rut depth in less than 1,000 load cycles. 30
Aggregate type influences APA test results as indicated by the ANOVA. Chert gravel mixtures 31
rutted significantly more quickly than mixtures containing granite or limestone aggregate. The 32
mixtures containing limestone aggregate performed the best of those tested. 33
Chert gravel aggregate had the lowest texture indices according to AIMS testing. Aggregate 34
texture is a better indicator than angularity of rutting resistance according to the Pearson’s 35
product moment analysis of APA test results. 36
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 17
The APA test results were not significantly influenced by maximum aggregate size for the 1
mixtures used in this study according to the ANOVA. 2
The APA test results were not significantly influenced by the relative gradation of the mixtures 3
for those used in this study according to the ANOVA. 4
The APA test using a hose pressure of 250 psi rapidly damages HMA specimens. Most mixtures 5
reached the terminal rut depth of 12 mm before 8,000 cycles were applied. 6
None of the mixtures tested in this study had less than 6 mm rut depth after 8,000 cycles in the 7
APA. The number of load cycles during the test will likely have to be reduced to produce 8
criterion that will not eliminate a majority of mixtures. 9
RECOMMENDATIONS 10
Based on the results of this study, a preliminary criterion of less than 10 mm rut depth after 4,000 APA 11
load cycles using 250 psi tire pressure is recommended for accepting HMA for high tire pressure aircraft 12
during mix design. This recommendation is limited to the materials used in this study. Further testing 13
should include more aggregate types representative of all regions of the United States. Additional binder 14
types, specifically modified binders, should also be included in future studies. Further, mixtures with 15
proven successful field performance should be evaluated with this criterion, along with mixtures that have 16
shown to rut easily. Further investigations should also determine the correlation of the APA test results 17
with actual in-service pavements. Finally, the repeatability of the APA at high pressures should be 18
investigated to develop tolerances for agency specifications. 19
ACKNOWLEDGEMENTS 20
The authors would like to thank the FAA William J. Hughes Technical Center for sponsoring this work 21
and Mr. Jeff Stein and Mr. Matt Wilson of the FAA for conducting APA testing. The authors would also 22
like to thank Dr. Eyad Masad, Texas A&M University, for his contributions on AIMS testing. Permission 23
was granted to publish this paper by the Director, Geotechnical and Structures Laboratory, ERDC. 24
REFERENCES 25
1. “Item P-401 Plant Mix Bituminous Pavements (Superpave),” Engineering Brief No. 59A, U.S. 26 Department of Transportation, Federal Aviation Administration, 2006. 27
2. Zhang, J., Brown, E.R., Kandhal, P.S., and West, R., “An Overview of Fundamental and Simulative 28 Performance Tests for Hot Mix Asphalt,” Journal of ASTM International, May 2005, Vol. 2, No. 5, 29 pp. 1-15. 30
3. Williams, R.C., Hill, D.W., and Rottermond, M.P., “Utilization of an Asphalt Pavement Analyzer for 31 Hot Mix Asphalt Laboratory Mix Design,” Journal of ASTM International, May 2005, Vol. 2, No. 5, 32 pp. 16-40. 33
4. Williams, R.C. and Prowell, B.D., “Comparison of Laboratory Wheel-Tracking Test Results with 34 WesTrack Performance,” Transportation Research Record 1681, Transportation Research Board, 35 National Research Council, November 1999, pp. 121-128. 36
5. West, R.C., Page, G.C., and Murphy, K.H., “Evaluation of the Loaded Wheel Tester,” Research 37 Report FL/DOT/SMO/91-391, Florida Department of Transportation, December 1991. 38
6. Lai, J.S., “Evaluation of Rutting Characteristics of Asphalt Mixes Using the Loaded Wheel Tester,” 39 Project No. 8609, Georgia Department of Transportation, December 1986. 40
TRB 2012 Annual Meeting Paper revised from original submittal.
Rushing, Little, and Garg 18
7. Choubane, B., Page, G.C., and Musselman, J.A., “Suitability of Asphalt Pavement Analyzer for 1 Predicting Pavement Rutting,” Transportation Research Record 1723, Transportation Research 2 Board, National Research Council, 2000, pp. 107-115. 3
8. Miller, T., Ksaibati, K., and Farrar, M., “Utilizing the Georgia Loaded Wheel Tester to Predict 4 Rutting,” Presented at the 74
th Annual meeting of the Transportation Research Board, Washington, 5
D.C., January 22-28, 1995. 6 9. WesTrack Forensic Team, “Performance of Coarse-Graded Mixes at WesTrack – Premature Rutting,” 7
Final Report, June 1998. 8 10. Rushing, J.F., “Development of Criteria for Using the Superpave Gyratory Compactor to Design 9
Airport Pavement Mixtures,” Master’s Thesis, Mississippi State University, Mississippi State, MS, 10 2009. 11
11. “Standards for Specifying Construction of Airports,” Advisory Circular 150/5370 10 E, U.S. 12 Department of Transportation, Federal Aviation Administration, 2009. 13
12. Kandhal, P., Motter, J., and Khatri, M., “Evaluation of Particle Shape and Texture: Manufactured 14 Versus Natural Sands,” NCAT 91-03, National Center for Asphalt Technology, Auburn University, 15 Auburn, AL, 1991. 16
13. Masad, E., Al-Rousan, T., Button, J., Little, D., and Tutumluer, E. (2005). Test Methods for 17 Characterizing Aggregate Shape, Texture and Angularity, NCHRP 4-30A Final Report, Report 18 Number 555, National Cooperative Highway Research Program, National Research Council, 19 Washington, D.C. 20
14. Asphalt Institute, Mix Design Methods for Asphalt, 6th ed., MS-02, Asphalt Institute. Lexington, KY, 21 1997. 22
15. Button, J.W, Perdomo, D., and Lytton, R.L., “Influence of Aggregate on Rutting in Asphalt Concrete 23 Pavements,” Transportation Research Record 1259, TRB, National Research Council, Washington, 24 D.C., 1990, pp. 141-152. 25
16. Ahlrich, R.C., “Influence of Aggregate Gradation and Particle Shape/Texture on Permanent 26 Deformation of Hot Mix Asphalt Pavements,” Technical Report GL-96-1, U.S. Army Engineer 27 Waterways Experiment Station, Vicksburg, MS, 1996. 28
17. Chou, Y.T., “Analysis of Permanent Deformations of Flexible Airport Pavements,” WES TR S-77-8, 29 U.S. Army Waterways Experiment Station, Vicksburg, MS, 1977. 30
18. Zhang, J., Cooley, Jr., L.A., and Kandhal, P.S., “Comparison of Fundamental and Simulative Test 31 Methods for Evaluating Permanent Deformation of Hot Mix Asphalt,” Transportation Research 32 Record 1789, TRB, National Research Council, Washington, D.C., 2002, pp. 90-100. 33 34 35
36
TRB 2012 Annual Meeting Paper revised from original submittal.